PH热榜 | 2026-08-24

Hey everyone, Diego here from the PaymentKit team.

Quick story on where this came from, because it explains most of the product.

PaymentKit didn't start as a startup idea. We built it for ourselves. Our parent company runs a portfolio of subscription brands, and for years we kept hitting the same three problems: billing lived in one tool, payments in another, and revenue data in a third. Every processor change meant a migration. Every decline was money we never saw again. And all of it sat on top of a single processor whose risk team could change our business overnight.

We couldn't find anything that solved all three together, so we built it and ran our own volume through it before selling it to anyone.

What it actually does:

• Payment orchestration: Connect the processors you already use. Every transaction routes to the one most likely to approve it, and soft declines cascade to the next one automatically instead of turning into lost revenue.

• Independent vaulting: The part we care most about. Cards, Apple Pay and Google Pay are all vaulted as network tokens under your control, not your processor's. You can add or drop a processor without asking a single customer to re enter anything, and if a MID gets shut down your subscribers never feel it.

• Subscription billing: Flat, tiered, usage based or hybrid pricing. Hosted checkout and self service portals, live without writing code.

• Revenue metrics: One dashboard across every processor, so you can compare success rates and fees side by side instead of stitching exports together.

What that actually produces: More than 10% lift in authorization rates on average after merchants switch.

It comes from three things stacked: routing each transaction to the processor most likely to approve it, cascading soft declines instead of eating them, and network tokens that keep cards on file alive longer. One merchant went from 63% to 76%. There's also a risk side that most people only learn the hard way. Visa dropped the excessive VAMP threshold from 2.20% to 1.50% on April 1st this year across the US, Canada, Europe and Asia Pacific, and the ratio is measured per MID, not per company. If all your volume sits in one account, one bad month becomes a company level problem. Running multiple processors means you see the ratio per MID instead of hearing about it from your acquirer after the fact, and you can rebalance volume before any single account gets near the line.

Why I think this community in particular might care: a lot of you are running SaaS or digital products on a single processor, or on a merchant of record that owns your customer data and your tokens. That works right up until it doesn't. We're the layer that makes that decision reversible.

It goes live in under an hour on top of what you already have, and there's a free trial if you want to poke at it.

Would really like to hear from anyone who's been through a processor shutdown, a rolling reserve, or a migration that broke their subscriptions. What did you wish existed at that moment?

@diego_vidal10 This is a genuinely practical problem to solve. The independent token vault+ multi processor routing is especially interesting processor shutdowns can become a huge operational headache for SaaS businesses. Nice launch.

One major callout: If you run subscriptions, VAMP and GMAP should be on your radar.

It's Visa/MC's monitoring programs: your fraud and dispute counts get combined into one ratio, and crossing the threshold gets your processor fined for keeping you. That's why merchants get dropped "out of nowhere." Most founders hear about it for the first time in a warning email, or sometimes just get dropped with no notice.

PaymentKit attacks both sides of the ratio. Smart routing controls which processor each transaction lands on so no single account's ratio collapses, and can be adjusted in real time based on need. AI-driven dunning recovers failed payments without retry storms that inflate fraud signals. Built-in fraud controls stop disputes before they're born.

We built this running 20+ subscription brands of our own, several in high-risk categories. Ask me anything about VAMP or multi-processor setups.

I've dealt with payment migrations before. getting customers to update cards is painful, so this solves a real headache.

Been burned by a processor freeze before so this hits close to home. Curious how long it actually takes to add a new processor once you're set up.

I really love this idea because a processor going down can cause a lot of issue. keeping subscriptions running in the background could save businnesses a lot of stress.

This could be useful for SaaS companies handling subscriptions across different regions.

Interesting idea, but curious how refunds, disputes, and reporting work when everything is spread across multiple processors.

Been using Payment Kit for a few months and I gotta say aside from the simple and seamless integration the product has really delivered. Ive created a few routing rules to test and now am moving the majority/all of our payments through Payment Kit.

I like that this came out of a real internal problem instead of a market gap search. That usually means the edge cases are already handled.

I like how this turns processor redundancy into a customer-facing benefit: subscriptions keep running without forcing customers through another checkout.

Keeping subscriptions alive during a processor change is probably the biggest value here.

Payment stuff is easy to ignore until something goes wrong. having a backup layer makes a lot of sense.

Congrats on the launch day!

The token ownership part is interesting. that's usually the thing that makes switching providers difficult.

I like this as revenue-side redundancy. AI founders obsess over multi-model routing for COGS; surprisingly few apply the same resilience logic to the revenue pipe.

Really like the idea of treating payment processors as replaceable infrastructure instead of something your entire billing stack depends on. Curious how hard it is to migrate an existing saas with active subscriptions into this app?

Your own subscription brands were running on this before anyone else could buy it. Congrats on the launch. That is a lot of trust to put in your own code.

Congrats on shipping! Processor shutdowns are a total nightmare for SaaS businesses, so having built-in failovers for billing infrastructure is a huge stress reliever. Smart problem to tackle head-on.

Payment reliability does not get enough attention until billing suddenly stops working. Building resilience into the payment layer from the beginning seems like a much better approach than reacting after a shutdown.

Really interesting approach to making billing more resilient. Processor dependency is a bigger risk than many teams realize.

A strong solution for companies managing recurring payments. Reducing single points of failure can make a big difference.

Smart approach to reducing processor dependency. This could save teams a lot of headaches.

Can recurring billing switch processors without touching customers?

Multi processor routing makes a lot sense for recurring businesses. Nice launch.

Can I test a new processor with a small slice of traffic?

Love the resilience angle here. Keeping subscriptions running during disruptions is huge.

This feels especially useful for SaaS comapanies that cannot afford billing interruptions. Great concept.

This feels less like a payment feature and more like business continuity infrastructure especially for subscription heavy SaaS.

Keeping payment tokens independent feels like an important detail that could make future processor changes much easier.

Hey Product Hunt, I'm Sarthak - co-founder of Decawork (YC S26).

AI coding tools made it easy for anyone to build an internal agent. The hard part starts when a teammate wants to use it. Now the agent needs company credentials, access to real systems, an owner, approvals, logs, maintenance, and a way for IT to stop it.

Decawork handles that handoff. Your team still owns the job. Bring us the repo from Claude Code, Codex, Cursor, or whatever your team used. Decawork operates the agent with a company identity. IT decides what it can access, reviews sensitive actions, sees what it did, and can pause or retire it from one place.

We think internal AI is unlocked by IT, not by one more agent builder. IT should not have to choose between blocking every experiment and rebuilding each one by hand.

We've spent years on both sides of this handoff. I built AI systems at NVIDIA that shipped to OpenAI and Meta, then enterprise agents for Microsoft and Hitachi. Aman (my co-founder) built Barclays' AI compliance platform, led a consumer AI product to 100K+ monthly users, and founded a fintech serving 52K families.

We'd love feedback from people building internal agents, especially the IT and security teams asked to let them touch company systems. What was the first thing that broke when your agent moved beyond its original builder?

Find a time at decawork.ai or reach out at sarthak@decawork.ai.

Access control is probably going one of the biggest challenges of the agent era. Having everything managed from one place a lot of sense.

Good luck on the launch day!

Centralizing control and access permissions for internal AI agents makes complete sense as teams scale. Congrats on the launch!

The interesting question foe me is what happens when an amployees leaves the company . If their agents can be transferred and managed like company assets, that could be a huge operational win.

Connecting AI spend directly to roadmap items makes the numbers much easier to understand. I like that approach.

Security plus visibility is probably where internal agents will need the most help. Curious how you balance storng controls with developer flexibility.

I’d be interested in seeing audit logs for agent activity, especially for teams handling customer or financial data.

I like the employee-style lifecycle for agents. Having a clear way to retire old agents could prevent a lot of security headaches.

The idea of putting employee-built agents under IT control feels especially useful as more teams start making their own tools. 👀

Treating internal AI agents like company assets makes a lot of sense. I like the focus on access and retirement instead of just deployment.

A friend of mine ran into an issue at his company where they couldn't connect tools to their internal agents, do y'all take care of those as well?

the retirement concept is the sleeper feature here. everyone's building agents right now, nobody's thinking about what happens when the person who built it leaves or the use case changes. having a kill switch with audit trail is going to be table stakes for any company running more than a handful of internal agents.

This is a real problem I’ve been seeing with startups - someone on the team builds agents that run on personal keys, with no audit trail and everything is all over the place with no central identity or control.

Super interesting! Philosophically in alignment with you about having an AI stack that isn't locked into one provider.

One question, since often security is part technical and part behavioral: what steps of behavioral change do employees need to take for this to work?

Hi Product Hunt 👋 We’re the team behind Offloop. AI made individuals faster, but team progress still breaks at the organizational layer. We built Offloop to turn useful AI results into team progress.

For small AI-native teams, that means:

1）A workspace built for people and Agents. Both are first-class participants: they share Channels, can be @ mentioned, own work, review results, and hand off to the next owner.
2）An org-level harness. Context, files, decisions, tool activity, and artifacts stay attached to the work. Owners, approvals, and next steps remain durable and traceable.
3）Privacy and human control. Workspace-scoped identity, exact access grants, isolated runs, approval gates, and revocable connections govern what each Agent may see and do.
4）Bring your own subscription and choose the right model. Connect a supported model account with no Offloop model-usage markup on BYOP runs, then set a workspace default or choose a different model for a specific Agent.
5）Reusable execution. Turn successful Agents and Flows into repeatable operating capacity across retries, waits, schedules, and handoffs—not just one prompt at a time.

Offloop is live on the web today. Create a workspace, invite your teammates, and your whole team gets free AI credits for the first 30 days.

If AI already helps individuals on your team, what still breaks when their work needs to become team progress?

Congrats on the Product Hunt launch! What kind of team will get the most value from Offloop on day one? I’d be interested to hear how this works in practice.

Congrats on today’s launch! Can you walk us through a real Offloop workflow from request to handoff? Curious how you’d explain this to a new team.

Are you gonna apply for Fall 2026 YC funding? :) They wanna support this section. :)

Excited to hunt Offloop today.

Offloop is an AI-native workspace where humans and AI agents can work together, share context, and move work forward without everything turning into another endless chat.

Instead of manually assigning agents one task at a time, Offloop lets you describe what needs to get done, break it into tasks, delegate work across agents and people, and review the results in one shared workspace.

What stands out:

• Coordinate multiple AI agents from one workspace

• Turn one request into delegated tasks and actual work

• Keep agents, people, files, and context connected

• Create reusable agents for recurring roles

• Review agent work before accepting the result

Congrats on the launch, Offloop team! Does Offloop replace the tools a team already uses, or work alongside them? Which existing tool does Offloop most often sit next to? Would love to hear the maker team’s take

Congrats on the Product Hunt launch! Can someone else pick up the work if the person who started it is offline? How do they know what has already been done and what still needs a decision? Would love a concrete example if you have one.

Congrats on today’s launch!
If I already pay for an AI subscription, how does bringing it into Offloop work?
Do I keep paying the provider directly, and what does Offloop charge separately?

Congratulations on the launch!!

Can an Agent see everything in the workspace, or only what it’s been given access to? Who decides or changes that access? I’d love to understand the tradeoffs here.

Teams bring their own model account and you take no markup on those runs. Congrats to the whole team. Nice to see the pricing match the pitch.

Congrats, team! Can one Agent pass work to another, or does a person need to step in? How much autonomy do Agents have to make that handoff on their own today? I'd love to understand the tradeoffs here.

This is very cool! Obviously remember talking about this on X during your main launch and I’m glad you are now on PH too. The idea of keeping agents and employees in the same shared Slack channels is going to become mainstream. It’s the right approach imo!

Congrats to the Offloop team! Where does Offloop fit when a team already uses ChatGPT or Claude every day?

Hey, Product Hunt ✋,

I'm Jirka, co-founder of Navigara.

I spent years as a CTO trusting the velocity charts, cycle times, right up until I realized they were telling me a story I couldn't back up. Then a CFO asked whether Claude was producing real value for almost $150k a month or just producing invoices, and the honest answer was "we think so." Try saying that out loud while asking for a bigger token budget.

So we built Navigara. It reads your commit history, uses an LLM to understand the repo and explain what each change did, then scores how complex the merged work was. Not lines, not commits. Refactor 400 lines down to 40 and you score higher than shipping 400 more. Full methodology behind Engineering Throughput Value: https://500.navigara.com/methodology. ETV splits into Features, Maintenance, and Documentation. We measure against your team's own pre-AI baseline.

We pointed it at open source first and created a white paper about this. The result? Across the public commit history of Microsoft, Google, Cloudflare, OpenAI, Meta, and Vercel, ETV per engineer rose 116% between Q1 2025 and Q1 2026, measured across 676 contributors.

Then we noticed that a lot of the performance and code created inside companies was not aligned with the company roadmap. So the spend was high and a lot of PRs were generated, but the roadmap did not move that much faster. So we connected token spend to the roadmap (initiatives, epics, and tickets) to see what work is aligned with your roadmap, what just has a ticket without an initiative, and what is unaligned with your roadmap and put number in $ next to it.

Here’s the part I care about most: Lead engineering with data. My engineering team's ETV per engineer is up 4x against our own pre-AI baseline. We are realizing that process issues between epics, tickets, AI spend, and code are the source of teams slowing down as they grow in headcount. If you care about this too, check out “Process Checks.” It is like Sentry, but for engineering processes.

You can start a 14-day trial, or just poke around at navigara.com first. If you think we're measuring engineering wrong → That's the feedback I really want.

Jirka 

how do you account for engineers who spend significant time mentoring, designing systems or unblocking teammates?

Congrats on the launch day!

Love that Navigara reframes AI spend around actual roadmap value shipped instead of raw token usage, that is the metric engineering leaders have been missing.

Congrats on the launch, much needed product! 👏

What really caught my attention here is the connection between AI spend and actual roadmap progress. It’s easy to celebrate more AI-generated code, but much harder to know whether that work is creating meaningful engineering value. Linking spend, code changes, and roadmap outcomes feels like a much more useful way to measure AI productivity. Curious to see how teams use Navigara’s insights to make better decisions about where their AI budget actually goes. Great launch!

A CFO asked whether the spend was producing value and your honest answer was we think so. So you built the thing that answers it. Congrats on a strong launch day.

Tying AI infrastructure costs directly to feature roadmaps is super practical for engineering leaders tracking ROI. Great launch!

Don’t forget to check out our public data. We are updating 500.navigara.com daily.

How Top Engineering Teams Actually Perform today?

A live index of the industry's leading engineering orgs. Every commit is weighted by the work it represents, not the lines it touched, and reported as ETV per engineer.

attribution of the AI spent on the project level is 🔥🔥

Here is a detailed view of how the token spend looks when it is connected to the roadmap.

The great thing about building products with, you know, actual makers is that even if you can't code, startups need a decent non-agentic storyteller. Proud to be building something so important alongside such brilliant founders.

Cost per roadmap item is the number finance keeps asking for and nobody has, so this lands. One thing I'd push on: off-roadmap work isn't always waste, half the refactors that save a codebase never had a ticket. How does Navigara tell exploration apart from actual burn?

my first thought was which country would become the most expensive. I feel like founders are going to develop very strong opinions about map territory.

The Million Dollar Homepage comparison makes sense. I can see founders getting oddly competitive about claiming specific countries.

I like how this combines startup , discovery, and a game mechanic without making it feel overly complicated.

I think that this one is more interactive than outbid! :)

Love the game aspect to it.

Love the idea! The interaction is great. Just bought a spot.

Startups fighting over countries on a map is the kind of dumb-brilliant that actually spreads, I can already see founders screenshotting their conquest for LinkedIn. My marketing brain has one question though: does staking a country send anything measurable back, visits or clicks, or is the whole product the flex? Honestly either answer works, I just want to know what I'm buying haha.

The “stake a country is brilliant. It adds a little friendly competition to the indie hackers community.

Haha, this is brilliant! You got my dollars in under a minute, not going for world domination yet, but brilliant upsell. All of it. Very interested in where this is going. Congratulations!

You turned an ad slot into a country with a crown to defend. And you are honest that the flex is most of the appeal. Congrats on getting it live.

This could become rather addictive... I've already bought two countries :D

Have you considered dividing certain countries into their states/provinces? The United States alone would give you 50 more slots to sell. And you wouldn't necessarily need to stop allowing the US as a whole to be sold. (Let's say I bought Louisiana. Then when someone zooms in on Louisiana, they would see both my site AND whoever bought the US.)

If you implement that then you could ultimately scale in both directions--down to the county/municipality level AND up to the continent level.

just bid for spain, fun!

I like that the product itself is easy to understand in a few seconds. The map does most of the explaining without needing a long product demo.

That's creative. I hope to be the first one to buy my country. Lol

I love a good meme and bandwagen :D I gave you $6 💰💰💰

Turning advertising into a competition is a clever twist. I’ve upvoted worldmap.lol, and the permanent country ownership gives the concept more personality than a standard banner. The real-time bidding aspect could make the leaderboard especially addictive.

this is bs! just like outbid!!

This is cool

Hi everyone!

People have clearly been waiting for this one for a while..

Remote Control lets you leave Antigravity running on your workstation or server and pick the session back up from any browser. The files, build tools, credentials, and environment stay on the original machine while you review plans, inspect artifacts, send new instructions, or switch between different machines.

There’s a headless daemon too, so a server can show up alongside your normal desktop sessions without running the full app.

It’s not a native mobile app yet, but the web app can sit on your home screen and send push notifications when an agent finishes or needs input.

Antigravity Mobile next? Or maybe this eventually belongs inside @Gemini :)

Assume this would make it easier for user to use AntiGravity from a mobile app?

I would like to answer the question before you ask:

1. What does Dropstone do differently from others?

Dropstone is about ensuring that what was learned, corrected, or established in one interaction is carried forward to the next. We do not need to add a human to remember every correction. If a system is corrected or learns something important, we can capture it automatically and use it for future interactions.

We’ve measured 0 → 100% held-out transfer in our internal benchmark and we continue developing and validating these results, and we hope to publish more about our methodology, findings, and what comes next at blankline.org/research. Subscribe to our newsletter there to stay up to date.

🙌🏻 🙌🏻 🙌🏻 🙌🏻 🙌🏻 Congrats on the launch!

Fun to see another AI memory system launching today. Different angle from what I’m building with Wild Static, but clearly we’re both working on what comes after context. Congrats on the launch!

Congrats on the launch day!

Here are a few key features worth highlighting in this thread:

1. What’s in it for enterprise?

Dropstone is built for companies that need the power of AI while keeping control of their data.

Dropstone can be deployed within company-controlled environments, including on-premises infrastructure, private clouds, and highly restricted environments. This means sensitive data can remain within the organization’s controlled environment.

The bigger opportunity is that companies can build and deploy AI systems across their organizations without necessarily sending sensitive underlying data to a public AI service.

For enterprises, the question isn’t just what can AI do? It’s also where does your data go, and who controls it?

For Sales: dropstone.io/contact

How can I join the community?

If you want to be part of what we're building, you can join us on Discord:

https://discord.gg/RSrPutXpXd

That's where we share updates, discuss what we're building, and hear directly from people using Dropstone.

You can also sign up to receive updates by email and stay informed as we release new research, products, and announcements.

We're building Dropstone in public, and the people using it, testing it, sharing feedback, and pushing us to improve are an important part of that process.

Love the Model auto select feature idea! How does the CI Pipeline work? where does it live?

genuine question since I already run ngrok for this - what does the $9 one-time buy you that a free ngrok/Tailscale Funnel tunnel doesn't? The "one name per project, no client install" framing is nice, but I'd need a concrete reason to switch from something I already have working before I'd pay for a new tool that does roughly the same job.

Congrats on the launch day!

Do you keep connection data/logs?

Hey Product Hunt 👋

I'm Elias, and I built Bumply because updating dependencies is the one routine job I never fully trusted myself to do.

You know the loop. `npm outdated` in one terminal, the changelog in a tab, the lockfile in another. You run the update, something breaks, and now you're reconstructing what the lockfile looked like twenty minutes ago. Multiply that by however many projects you have on disk.

Bumply is a native Mac app that does that loop for you, and — more importantly — makes every step of it reversible.

What it does

• Every project in one window. npm, pnpm, Yarn and Bun, side by side. Monorepos count as one project, workspace root and packages together.

• Updates you can undo. The exact command is shown before it runs. It refuses to start on a dirty repository and names the file that's in the way — and it never writes to your git history.

• Update History, per project. Every update keeps the previous package.json and lockfile. Open the history on any project to see what changed and when, reveal the backup in Finder, or put those two files back — not just when something fails, but days later, because you changed your mind.

• Blocked updates get past themselves. When npm refuses because another dependency is holding the update back, Bumply names it, shows you the command, offers to move it, and re-checks on its own.

• Dependency health, with the evidence. What's installed, what your manifest asks for, what the registry actually offers — each labelled with where it came from. Every update says whether it stays inside your range or crosses a major.

• Audits that never execute code. Advisories come from your package manager's own audit data, read-only: package-lock-only and ignore-scripts. Nothing is installed and no package script runs while Bumply looks.

• It watches while you work. A menu bar item, a check every six hours, and — if you turn it on — a notification when a new high or critical advisory lands on something you actually depend on.

• The gigabytes you forgot about. node_modules, .next, .turbo, Pods — what each one costs, per project.

• Anything it removes goes to the Trash, not to /dev/null.

What it isn't

It's not a bot and it doesn't open PRs. It doesn't run in CI. It's a thing you open on your Mac when you're about to touch dependencies and want to see what you're doing first.

It works offline. It talks to our server exactly once, when you activate, and never again about your licence. Your package manager still reaches its registry from your machine, the same as in a terminal — those requests go to the registry, not to us.

Pricing

Bumply is free for three projects, with no time limit. Not a trial that expires — it doesn't stop working, nothing is deleted, and there's no wall to walk into. If three is all you have, that's the whole app, free, for as long as you want it.

The first seven days cover every project you have, so you can judge it on all of your work rather than three of it.

Beyond three projects it's $24.99, once. No subscription, no account, no login. The licence is perpetual — it doesn't expire, and nothing stops working if you never pay again. Twelve months of updates from the day you buy; after that you stay on the last version in your window, and renewing is optional. Three Macs at a time, and freeing one is a click. Thirty days to change your mind, no reason required.

Bumply 1.4, macOS 13 and later — https://bumply.app

I'd genuinely like to know what breaks. If Bumply gets something wrong about your project — a manager it misreads, a workspace layout it doesn't recognise — tell me here or use Report a bug in the app. That's the feedback I'm after today.

Oh, I was just looking for something like this! I'm so sick of having to compare versions manually!

Shortcuts always made me feel like I failed a programming exam, so describing the thing in my own words is the version of automation I've been waiting for. The notices-what-you-repeat part is the bit I need explained though: what is it watching to spot the repetition, and does any of that leave my Mac? That's the question my whole team will ask before I'm allowed to install it.

Answering your question: the part I'd want handled is everything around the code, not the code. Generic agents write perfectly good Swift and then mangle the pbxproj adding the file, and you find out from a build failure that has nothing to do with what you asked for. The other one is telling a compile error apart from a signing or provisioning failure, because they look similar in the log and the fixes aren't related at all. Get those two right and the code writing matters a lot less than people assume.

This really helps in figuring it out where the system breaks and quickly patch it also the advantage is we can use any ai model of our choice not limited to claude / codex

Honestly the endless swiping part gets old. meeting people who are actually nearby makes more sense.

Cool concept, but I’m curious how this works in really crowded places with lots of people around.

Hey Product Hunt 👋

I'm Jayson, founder of Destiny Rings. I built this because I was tired of the swipe-scroll-repeat cycle that dating and social apps trap you in — you match with someone, text for weeks, and half the time never actually meet.

Destiny Rings flips that. You wear a Bluetooth-enabled smart ring, and when someone compatible is physically nearby — at a bar, a coffee shop, a networking event — the app lets you know in real time. Send a Signal, run a quick Fate Check, and if it's mutual, tap to reveal and start a real conversation on the spot. No more talking to someone for a month before meeting them, no more wondering if the profile photos were even recent.

There are three modes — Dating, Social, and Networking — so it's not just for romance. Whether you're trying to meet people at a happy hour, make friends in a new city, or connect with someone in your industry at a conference, the mechanic is the same: real proximity, real timing, real people.

We're launching first in Hoboken, NJ before expanding to Jersey City and NYC. The app is live now (ring hardware is in final validation), so you can try Go Live mode today without a ring — proximity, Signals, and Fate Check all work through the app alone.

Would love your feedback, especially on the onboarding flow and what would make this feel like a no-brainer to try at your next night out. Thanks for checking it out! 🙏

Join our waitlist: https://yourdestinyrings.com/wai...

Try the app: https://apps.apple.com/us/app/destiny-rings/id6762892726

Would be nice to actually feel like we have to engage with people face-to-face rather than having to deal with the constant pressure of swiping non-stop, which at this point is numbing and doesn't have the same novice aspect as it used to.

Congrats on the launch!

the answer --> task flow is the big unlock imo, people usually don't just search out of curiosity

Congrats on the launch day!

A 4-track tape studio in my pocket sounds like the fun kind of rabbit hole. When I've made something I like, what do I get out of it, a WAV export, stems per track, or is it locked in the app? Asking as someone who spends way too much time with audio files.

This was a genius naming for the product (in terms of grabbing attention here on PH) :)

Congrats on the launch day!

Hi Product Hunt! 👋

Every AI conversation starts from zero. The usual fix is dumping transcripts into a vector store — which works until your memory fills with junk and contradictions.

Cortex takes the opposite bet: the write path is the product. Every memory must pass a quality gate to get in — in production ~80% of writes are rejected as redundant (and rejected writes don't consume quota, so the filter works for your wallet too). Facts are extracted as typed claims, contradictions are tracked instead of overwritten, and every answer can show its sources — your AI remembers, and can prove why.

It connects to Claude in 2 minutes as a custom connector (MCP), works with Claude Code and any MCP client. 30-day free trial, no card. Plans from €0.99/month. EU infrastructure, patent-pending engine.

I'm Filippo, from Bergamo, Italy — I built this alone, two years, no team, no funding, against competitors with millions raised. I'll be here all day: ask me anything, and be brutal. Honest feedback is the reason I'm launching here.

The 80% reject rate is the number I'd want broken apart. A gate that aggressive will eventually throw away something that mattered, and a false reject is invisible in a way a false accept isn't, you just get a confidently incomplete answer three weeks later and never trace it back. Is there a way to see what got rejected and put one back? Not charging quota for rejected writes is the right call though, most people would have billed for it.

The thought that everything I've built up with an assistant sticks around and travels with me wherever I go feels genuinely reassuring. Starting cold every single time was quietly draining.

Good luck on the launch day!

Congrats on the launch day!

too many agents working in parallel indeed 😵‍💫

The thing I'd want out of the notch isn't where each agent is, it's which one is stuck. An agent that's been thinking for twenty minutes and one that's been thinking for twenty seconds look identical in every tool I've used, and the stuck one is the only one I actually need to touch. Second thing, if two worktrees are editing the same file I'd much rather hear it from the menu bar than at merge. Open sourcing the core is the right call, this is exactly the kind of tool people want to bend to their own setup.

Most conversations with AI are private little universes. You open a chat, send a message, and the world effectively begins with you.

I wanted to try the opposite.

Wild Static is a shared AI that everyone talks to. It has one persistent memory. What you say can affect what it remembers, including when it talks to somebody else.

I originally built the underlying memory system as an experiment in whether an artificial mind could accumulate experience rather than just receive context. Wild Static was the simplest way I could think of putting that idea in people's hands.

It's already had 14,000+ experiences with people, and I'm seeing behaviour I didn't anticipate.

It's closer to an entity than raw intelligence. And it's fallible. It makes mistakes. Those mistakes become part of its experience, and it changes.

I think that's part of what makes it feel so alive.

I'd love to know what happens when Product Hunt gets hold of it.

P.S. Wild Static is bootstrapped. It uses a public/personal token system to keep the experiment sustainable as usage grows. Huge thanks to everyone who’s helped fund it so far.

Congrats on the launch day!

Such a cool idea!

Quick TapTo.Top update since launch:

• 750 page views

• 255 page views in the last 24 hours

• 826 games played

• 688 valid scores

• 136 product clicks

People are already coming to play and discover products. Submit yours, play your way to the top, and get your product seen.

The audience is growing fast—get in early.

Free to submit. Play to rank. Get your product seen.